Académique Documents
Professionnel Documents
Culture Documents
of Molten Polymers#
E. C. ACHILLEOS', G. GEORGIOU', and S . G. HATZIKIRIAKOS2*
'Departmentof Mathematics and StQtistics
P.O. Box 20537
University of Cyprus
1678 Nicosia, Cyprus
the production rates of commercially acceptable products are limited to the ones below these critical values.
In order to overcome these difficuties and to render
the processes economically feasible, processing aids
(PAS) are frequently used. PAS eliminate flow instabilities or postpone them to higher flow rates. The end
result is an increase of the productivity as well as an
energy cost reduction, while high product quality is
maintained.
The purpose of this review is to examine traditional
and new pmcessing aids that have been investigated
over the past two decades and to discuss their performance in melt hcture elimination, so as to elucidate the role of these materials in instability elimination
and polymer processability improvement. The effect of
PAS on the rheologid properties and the instability
(melt fracture] elimination is also discussed.The factors that determine the additive performance, such as
concentration, dispersion quality. and interactionsbetween additives, additive and polymer, and additive and
die surface, are also examined. The various mechanisms by which Merent types of processing aids help
eliminate the instabilities are reviewed as well. Section
2 discusses the various melt fi-acture phenomena in
The flow instabilities that occur in extrusion processes can be better explained by looking at the different flow regimes of the apparent flow curve (3.4,11,
12).This is a logarithmic plot of the wall shear stress
versus the apparent wall shear rate when a capillary
rheometer is used, or the plot of the pressure drop
. =32Q
y*
7rD3
where Q is the volumetric flow rate, and D is the diameter of the capillary. The wall shear stress, uw, is
defined as (13):
- A P - APITw
4L/D
m. 1. Qpicclrjbw curuefor UnearpolyethyIene acbuded in apiston driven cqpillnry tllustmting the various regions of instability.
8
appear on the exbudate surface. This instability is accompanied by a cllange in the slope of the flow curve,
as reported by many investigators (3.4,14-16).This
instability is observed up to the end point of the left
branch of the flow curve (often referred to as the low
flow rate branch) and is genereferred to as sharkskin or surface m e l t w e . A second instability occurs when the shear stress exceeds a second critical
value (crc2) and only within a certain range of apparent
flow rate values. It corresponds to the region separating the two branches of the flow curve, and is usually
referred to as stick-slipinstability or spurt phenomenon
or oscillating melt fracture. In this regime, the flow
ceases to be stable when a controlled rate capillary
rheometer is used. Instead, pressure, and thus shear
stress, oscillates between two extreme values and alternate distorted and smooth zones appear on the extrudate surface (3,4,9, 11, 12,17-19).Finally, at even
higher apparent shear rate values, the flow becomes
again stable, whereas the extrudate exhibits gross, irregular and chaotic distortions. This type of instability,
is referred to as gross meltfracture, and corresponds to
the right branch of the flow curve (1,3,4,11,12,
20,
21,22).
Thorough reviews on extrusion instabilities have
been published, and the mechanisms responsi%le for
these instabilities have been the subject of ongoing
debate (1,5,6,7.23. 24).Various polymer rheologid
properties, such as wall slip, melt compressibility,melt
viscoelasticity, non-monotonic shear stress-shear rate
law, viscous heating;, viscoelastic rupture, and local
stick-slip, which may be simultaneously involved, can
be held accountable (5,6,7,24. 25,26).One has to
distinguish between the mechanisms responsible for
each type of instability. For the case of sharkskin melt
fracture, there is general agreement about its location
of occurrence. It takes place at the exit of the die, either
because of rupture. as the boundary condition changes
from no-slip or partial slip to a shear-free one, which
results in a highly extensional flow (14,27,28,291,or
because of local stick-slip at the exit as a result of
polymer molecule disentanglement (30).On the other
hand, gross melt fr-acture occurs at the entrance of the
die because of the abrupt contraction (20,31).Since
this review focuses on the elimination and not on the
mechanisms that cause these instabilities, the latter
will not be discussed fiather.
2.2.
wdl slip
Wall slip usually accompanies the instability phenomena under discussion. It is, indeed, the phenomenon most thoroughly studied in the literature, both
experimentallyand computationally.
Sharkskh instability has been attributed by many
researchers b small d q m l w t s from the no-slipboundary condition, which am be accompanied by a sharp
change in the flow curve slope (3.4.11, 32.33).Some
experiments have indicated that sharkskin or surface
melt fi-acture is suppmsed byni
the adhesion
between the fluid and the wall (3,s).
whereas others
have shown that slip promotion suppresses the instability (28,35,36). Slip velocity increases when certain
additives are used and as a result shear stress is reduced, shifting the flow curve to higher shear rates
(37,381.
This is because both the slip velocity and the
critical shear stress for slip are strong functions of
the work of adhesion between polymer and interface
(37-40).It is the shear stress reduction, rather than
the slip velocity increase, that suppresses the instability at a particular flow rate, when additives are used.
Moreover, it is also the type of slip that causes sharkskin to occur. over clean swfaces, polymer slips locally
by a stick-slip mechanism that creates the well-known
schallamach waves similar to those in the sliding of
elastomers over metallic surfaces (41).Polymeric fluids
are not much Merent from elastomers at high enough
shear rates, since their elastic properties prevail in
short time scales.
At low shear rates, slip was speculated to be caused
by an adhesive failure, that is, stress-induced strain
detachment of polymer molecules form the wall (3.4,
1 1, 42,43).Experiments have shown that there is a
strong correlation between the work of adhesion of the
polymer/wall interface and the onset of slip (39,40)
and this finding implies adhesive failure. As discussed
above, experiments have indicated that the sharkskin
originates at the die exit region because of polymer
rupture, as a result of the high stresses that develop
around the die exit (14,27,44, 45).This mechanism
is certainly consistent with the violation of the no-slip
boundary condition. Gross melt fracture originates
near the die entry region, where unstable vortices appear and symmetry is lost (35,44, 4.6,47). Cohesive
Mure, that is sudden chain disentanglementand rupture of chains near the wall. can account for slip in
the gross melt fracture regime, and thus for the unstable and v i r t d l y plug-flow behavior (20,48). Overall, the experiments in the literature indicate the presence of weak slip (small deviations from the no-slip
conditions) in the sharkskin regime, which is attributed to adhesive failure and of strong slip in the grw
melt fi-acture regimes, which can be explained by cohesive failm.
Other experiments have shown that slip is not at all
present in the sharkskin regime, indicating that other
phenomena also contribute to this instability (30,35,
49,50).The stick-slip and the gross melt fracture instabilities have been modeled by many researchers
using a combination of a slip law with the small but
finite compressibility of polymers (51-56)or with a
viscoelastic model (57-60).Both compressibility and
viscoelasticity can account for the storage of elastic
energy that sustains the pressure and flow rate oscillations in the stick-slip regime. Melt compressibility in
the reservo&was shown to be a necessary (9)but not
sufficient condition (61)for the variations in the pressure and flow rates in the stick-slip regime. Cohesive
Mure at high shear mtes was attributed to hardening
of the polymer, which undergoes a relaxation transition from a fluid to a highly elastic solid-like state (47,
3.TRADITIONALPROCESSING AIDS
3.1 Flaomymers
Fluoropolymem have been traditionally used as processing aids in the extrusion of commercially important
polymers, including linear low-density polyethylene
(LLDPE), high-density polyethylene (HDPE), polypropylene (PP), copolymers of ethylene, propylene and
hmdiene (EDPM), and others. Low concentrations of
fluomelastomers (down to 500 ppm) in dispersion with
the processing polymer can significantly enhance product quality and processability, not only by eliminating surface defects, but also by reducing torque and
power requirements as well as die plate pressure (67,
68).Fluoroelastomers used as PAS include Dyneon's
DYNAMAR@,
DuPont's m N @ (a copolymer of vjnylidene fluoride, hexaethylene and/or tetrafluoroethylene). W o n t ' s TEFLON@,and DFL (dry film type).
An important effect that fluoropolymers have, when
used in dispersion with the processing polymers. is the
increase of slip velocity, which is accompanied with a
reduction of pressure. Thus, the apparent shear stress
needed to extrude the polymer at a particular flow
rate is much lower, or alternatively, an increase in the
throughput rate is obtained at a given pressure drop
(67, 68,70, 74). This is accompanied by a delay of the
sharkskin instability to higher shear stresses. Athey
et at (68)
found that when a version of ~
N (vinyli@
dene fluoride, hduoropropylene 60:40)was used in
dispersion with HDPE, LLDPE, and PP. the shear stress
was reduced and the flow curve was smed so that
the sharkskin instability was postponed to higher
shear rates. The reduction of shear stress took place
up to shear rates of 1000 s-l, therefore eliminating
only the sharkskin and the stick-slip instability but
not the gross melt fracture. Similar results were found
10
onsetof
SharkskinMF
onsetof
GrossMF
0.01
1
I0
100
lo00
loo00 lo0000
be optimized. include the dispersion quality. the particle size, and the interactions with other additives
present in the processing polymer. Generally. the dispersion quality improves when the additive has a small
particle size 6.e. 5 pm) and a homogeneous dispersion
is prepared with intensive mixing. A two-step procedure
in which a master batch is first prepared and pelletized,
and then mixed with the process polymer is an effective route for preparjng homogeneous dispersions (67,
74). h the case of W N @ . diluting the fluomelastomer
in acetone prior to mixing was found to be beneficial
(79). The interactions with other additives that might
be present need to be minimized if not eliminated completely (74, 80,81. 82). Wester et aL (74) found that
the presence of anti-block agents slow the conditioning
curve (time needed for steady state) and that higher
additive concentration is required in order to have the
same conditioning curve as in the case of no additive.
It is generally agreed that fluoropolymm when used
in dispersion migrate to the wall of the die and create
thin coatings (68, 74, 75. 76, 79). Generally, fluoropolymers are more polar than the process polymers so
they migrate more easily to the wall. They are deformable at the extrusion temperatures so they can easily
form a coating on the die wall (68).
The m t i o n of secondary flows as a result of sharp entrance effects,
which create eddies, can assistparticle migration to the
wall (79). As in the case of coated dies, the interfacial
conditions determine the molecular mechanisms that
can either promote or suppress slip,with subsequent
results on the shear rate reduction and/or instability
suppression. Experiments show that with the exception
of DFL (34)fluompolymers act as lubricants, promoting
11
mtsl.
slip. Slip promotion is attributed to the incompatibility of the host polymer with the fluoropolymer layer
(38,68,78).The experiments of Hatzikiriakos and
co-workers have shown that when fluoropolymers are
employed as PAS, the slip velocity increase is accompanied with a decrease of the critical stress for the onset of slip (34,37,38,70,77).
In addition, Anastasiades and Hap ** ' * ' (40)
a sessile drop method to measure the work of adhesion
between HDPE/LLDPE and clean as well as FE-coated
stainless steel. They found that VITOP and, more
profoundly, TEFLON@ reduce the work of adhesion between polyethylene and the surface. These findings
support the hypothesis that slip is a result of adhesive
failure, and that slip velocity and critical shear stress
for the onset of slip depend on the work of adhesion.
Slip models taking these phenomena into accounf have
been developed by Hatzikiriakos (38).Hill et d (39)and
Stewart (43).The fact that instability enhancement is
associated with the onset of slip for virgin resins is
somewhat inconsistent with the instability suppression effect that fluoropolymers have when they act as
lubricants. H
a
m and Dealy (34).hawever, point
out that the presence of either an adhesion or slip promoter can alter the variation of slip along the die, and
reduce the extensional strain upstream the die exit,
which according to Kurtz (14)is a determining factor
for the stresses that develop at the die exit and thus
for the occurrence of the instability.
12
c q i k r y expen-
that contain steruates, the die-conditioningphenomenon is observed, :similar to the one depicted in Rg. 3.
In several experimental runs, HatzWriak0s et al. (83)
showed that in capillary experiments with metallocene
long chainbranched LLDPEs that contain stearates. the
pressure initially iises and then starts dropping. After
s e v d consecutive capillary extrusions, i.e.. by reloading the reservoir of the rheometer, pressure attained
its steady state value, indicating that complete conditioning of the die surface has been achieved. Comparison of Merent resins has shown that the conditioning is longer as t h e amount of long chain branching is
increased. The critical shear stress for the onset of slip
for these resins was 0.05 MPa, significantly lower than
the one corresponclhg to linear low-density PE resins
(0.1 MPa). Furthermore. sharkskin and stick-slip instabilities were observed only in the transient experiments, whereas in steady state, only gross melt fi-acture is observed. Gross melt fracture, accompanied
with a change in the slope of the flow curve, was observed at 0.33 Mpa, which is similar to the shear stress
values observed for polymer having conventional PAs.
Overall, the processing behavior of metallocene polyoleflns was similar l o that of polyethylene when fluoropolymers are used. Since no other additive was detected in the metallocene resins, it was concluded that
calcium stearate, wfuch are present in concentrations
of about lo00 ppm, iTCt as processing aid with a mechanism similar to that discussed for fluoropolymers.The
fact that stearates do not, to our howledge, enhance
the processability of linear low-density polyethylenes
could possibly be explained by investigating possible
interactions between stearates and LDPE as well as
with metallocene potyoleflns having a small degree of
long chainbranching. More experimental evidence and
systematic studies far resins with various concentrations of stearates are needed for more conclusions to
be drawn.
h d s
Hydromrbonsblended with the process polymer can
also act as PAs. This 1:echnique might involve not only
dispersing a s d quantity of PA into a process polymer, but also polymer blending to ratios of the same
order of magnitude. Although the notion of a processing aid has a meaning at small concentrations (typically less than 1%)).blending at higher concentrations
might also improve processing. Examples from the lite r & . ~ are
~ included hiere.
Blending of polypropylene (PP) with polytethyl vinyl
acetate) (EVA) for flow instability reduction of PP was
examined by Montoya et aL (84)with capillary rheometer experiments. They tested blends of concentrations
ranging h m O?! to 180% and found that the critical
shear rate for the onsel of fi-acture increased with EVA
concentration. Mechanical property measurements
have shown that while pure EVA has completely elastomeric behavior, blends of 50%PP and higher exhibit
thermoplastic behavior with yield stress that increases
with PP content. It was thus concluded that 15/6-20%o
13
, # I ,
T=350C U D 4 0 D=0.762 mm
$3
cr
0.2
.E
c
-h
0.1 -
Q)
of PE {GRSN-7047)on theflow
curue of an FJP resin at 350C.
0.08
v v
=I
0.06
fn
x 0.04
2
(P
(I)
ll A v
I10 V
0.02
V
V
102
103
104
of the extrusion pressure and the elimination of sharkskin in a similar fashion as fluompolymersare used in
PE extrusion. It is noted that fluoropoymers, such as
TEFLON@and FEP, have been used effectively as PAs
in the extrusion of PEs (15,30, 34, 67,68,70,89).As
discussed above, it is expected that PE would tend to
migrate towards the die wall during extrusion, since
its viscosity is lower than that of the FEP. In fact, it is
not the viscosity but the work of adhesion that allows
this process (migrationand wall coating)to take place.
It is the energy required to remove the polymer from
the metal surface that determines which component
eventually coats the die wall. Polyethylene has greater
adhesion to metal surfaces than FEP resins, so it is
not surprising that during extrusion experiments the
metal die surface eventually becomes coated with a
very thin layer of polyethylene. Since there is very little
adhesion between FEP and polyethylene at sufficiently
high shear stress, it appears that FEP slips over the
thin polyethylene coating on the die wall.
This, then, raises the question as to why a very
small quantity of FEP resin, when h e l y dispersed in
polyethylene, will completely coat the die wall during
extrusion. The answer again lies in the relative values
of the work of adhesion. Although the work of adhesion of FEP to metal is less than the work of adhesion
of polyethylene to metal, the work of adhesion of polyethylene to FEP is much less than either of these.Thus,
if a very small FEP particle dispersed in polyethylene
comes into contact with the metal die surface during
extrusion, there will be very little force at the FEP/polyI
ethylene interface acting to pull the particle off the
0.20
T=325OC yA=104.2s-l
UD=40 D=.762 m m
2
z
0.16
a. Teflon FEP 4100
b. Teflon FEP 4100+0.1% PE
2
v
u)
u)
0.12
er
u)
tiia,
u)
0.08
er
(P
0.04
It
0
50 0
l a , , , , , II,,
50
,, ,
300
350
15
The use of several BN powders as processing additives in extrusion of metallocene linear low-density
polyethylene (m-LLDPE) and TEFLON@ FEP/PFA resins was experimentally investigated employing capillary rheometers, with either annular or cross-head
dies, and larger-scale extrusion equipment (90-95).
The effectiveness of the additive was found to depend
on the die geometry, the temperature, the dispersion
quality, and the BN concentration and type.
Capillary rheometer e x p e ~ e n t shave indicated that
BN has a strong impact as a processing additive when
16
#I
I ) I ,
smooth
a
Q
PE Exact 3128
5
0.1
-a
I
m.8. Cro~s-headdief
, 1 1 1 1
virgin resin
0.05% BN
,,I
wire coating.
17
0.01% BN
0.05% BN
8
1
0.9
0 0
A
0.8 0
8
102
V 8a!@?*
0.5% BN
6'1 1
il
103
18
Q- 11. Ewtnrdate samples to iUustrate the eff'ectof BN (ClF5)concentration on the extrude appeamnce of m - D P E Exac@31 28
obtained at 617 s-l arld 163C:1) pure resin:2) 0.02% BN: 3)0.1% BN;4)0.5% of BN (ClF5).
processing aid. W n g grounded polymer with a concentrated BN-grounded polymer master batch (i.e.
10% BN) in either a single-screw or twin extruder, and
obtaining the polymer material in a pelletized form resulted in more uniform dispersions and higher effectiveness in instability elimination than dry mixing of
the BN powder with ]polymerpellets. which resulted in
only minor instabilily suppression (31. 92-95). This
effect is similar to tttat obtained in the case of traditional fluoroelastomers. where the dispersion should
be prepared by means of melting a master batch first.
The effectivenessof' BN in eliminating the flow instabilities depends not only on the concentration of the
additive but also on the type of powder used. Several
BN powders were tested for their ef6ciency. The basic
powder was CTF5.which has an average particle size
of 5 bm. CTL40 is an agglomerated version of C F 5 ,
and CTUF contains higher concentration of boron oxide than CTF5. Agglomeration and presence of oxides
are two factors signifitmtly alecting the performance
of BN as a n effective processing aid. Table 2 summarizes the morphological characteristics of the all
powders used in literature and the maximum shear
rates achieved in capillary extrusion experiments of
m - D P E at 163OC, using a cross-head die (31.92-98).
Dispersions of CTFS,CTUF,and CTL40 were prepared
using a single-screw extruder, and dispersions of
BN427-BN431 were prepared with a twin extruder.
The concentration dependence of powder performance
and the optimum concentrations for each powder are
indicated. The best additives, CTF5 and BN431, were
the non-agglomerated powders, in the size range of
5-10 Fm. containing minimum amounts of h03(9297). Properly dispersed in the polymer, at optimum
concentrations, and with the preferable geometry
(cross-head die), these two additives can postpone the
onset of gross melt fracture to high shear rates (9401030 s-9, which are far beyond the value that gross
melt fracture appears in the virgin resin (500 s-l).
19
Table 1. MorphologicalCharacteristicsof the Various BN Powders Tested, and Additive Performance in Terms of CFtical Shear
Rate for the Onset of Melt Fracture in Capillary Rheometer Extrusion of m-LDPPE Using a Cross-head Die at l63'C.
BN
type
Approximate
Particle size
SEM (pm)
b 0 3
content
Agglomeration
("/I
Pure resin
10
Pure resin
CTF5
10
No
CTUF
Dispersion
concentration
(Yo of BN)
42
0.02
0.1
0.5
926
0.02
155
155
155
42
10
No
2%
0.1
0.5
CTL40
40
926
617
0.02
0.1
0.5
62
93
77
BN427
0.02
0.1
155
77
BN428
1.5
No
0.02
0.1
308
Yes (size>200)
0.02
0.1
155
155
0.02
0.1
124
155
0.02
865
1080
BN429
BN430
20
No
BN431
No
0.1
20
Maximum shear
rate for smooth
extrudate (s-')
124
Combinations of BN powders with fluoropolymer additives in order to further improve the processability of
the polymers were examined by Hatzikirialros and coworkers (91, 93. 95, 96,98). Experiments in capillary
rheometers with cross-head dies with PE Exact 3128
and PE Exceed 116 demonstrated that the combination of both additives results in greater processability
enhancement (91, 93, 96, 98). In the case of PE Exact
3128, addition of 0.05% TEFLON@ or BN shifts the
critical apparent shear rate for hcture from 50 s-l to
1300 s-l or 1800 s-l, respectively, whereas the combination of TEFLON@and BN shifts the critical shear
r a t e t 0 2 4 0 0 s - ~ , a s c a n b e s e e n i n11
~ . (93,95,96,
98). Similar results where obtained for the extrusion
of PE Exceed 116 (91, 96,98). Addition of 0.1% BN
shifted the uitical apparent shear rate for fi-acturefrom
110 s-l to 900 s-l, whereas combination of 0.1% BN
and 0.05% TEFLON@shifted it to 2000 s-l.When used
as the sole additive, BN had no or minimal effect on the
flow m e whereas combination of BN and TEFLON@
resulted in a significant shift of the flow curve. The
shear rates at which stable flow was achieved with
the combination of the two additives were well above
the point that gross melt fracture occurs.
was previously discussed,T E ~ N acts
@ as a slip
promoter, whereas BN does not. The fact that these two
additives have a different mechanism for suppressing
the instabilities explains why their combination results in further instability elimination, i.e. in steady
flows at higher shear rates, whereas increase of the
concentration of either one of the additives does not.
21
REFERENCES
1. C . J. S. Petrie and M. M. Denn, NChE J., 22,209
( 1976).
2. C. D.Han. Rhedogy in Polymer Processing,Academic
Press, New York (1976).
23
92. K. Yip, E. E. Rosenbaum, S. K. Randa, S. G. Hatzlldriakos, and C. W. Stewart, ANTEC, Soc. Plastics Engrs.,
Tech. Papers, 45, 1223-1227 (1999).
93. F. Yip, S. G. Hatzikiriakos, and T. M. Clere, Proceedings of SPE Polyole5ns 2000, 563-570, Houston, TX
(Feb. 27-Mar 1, 2000).
94.F. Yip, S. G. Hatzikiriakos, and T. M. Clere, ANTEC
'00,SOC.Plastics Eng'rs., Tech. Papers, 46, 2852
(2000).
95. F. Yip, S. G. Hatzikiriakos, and T. M. Clere, J. Vinyl &
Add&
Technology,6, 113 (2000).
96. M. Seth, S. G. Hatzlldrialros, and T. M. Clere, ANTEC
'2001, Soc. Plastics Engrs., Tech. Papers, 47, 2634,
(2001).
97. M. Seth,S. G. Hatzlldrialros, and T. Clere, Po@ Eng.
Sci, to appear (2001).
98. M. Seth,F. Yip, and S. G. Ha-os,
ANTEC '2001,
Soc. Plastics Engrs., Tech. Papers, 47, 2649, (2001).
99. R J. Binnington, G. J. Troop, and D. V. Boger, J. NonNewtonian Fluid Mech, 12,2555 (1973).
100. J. P. Tremblay, J. Non-NewtonianFluid Mech,52, 323
(1994).
101. F. Legrand and J. M. Piau, J. R h e o L , 42, 1389 (1998).
102. M. M. Britton, R W. Mair, R K. Labert, and P. T. Callaghan, J. RheoL,43(4), 897 (1999).
103. H. Munstedt, M. Schmidt, and E. Wassner, J. RheoL,
44,413 (2000)
(2000).
24